Constraining Dark Energy with the Supernova Legacy SurveyMark Sullivan University of Toronto

http://legacy.astro.utoronto.ca/http://cfht.hawaii.edu/SNLS/

Paris GroupReynald Pain, Pierre Astier, Julien Guy, Nicolas Regnault, Christophe Balland, Delphine Hardin, Jim Rich, + UKGemini PI: Isobel Hook, Richard McMahon, + USALBL: Saul Perlmutter, + CIT: Don NeillFull list of collaborators at: http://cfht.hawaii.edu/SNLS/ Victoria GroupChris Pritchet, Dave Balam, + Toronto GroupRay Carlberg, Alex Conley, Andy Howell, Kathy Perrett, Mark Sullivan The SNLS collaborationMarseille GroupStephane Basa, Dominique Fouchez, +

White DwarfSNe Ia are thermonuclear explosions of C-O white dwarf starsStandard nuclear physicsBright: 10 billion sunsStandardizable: 7% calibrationBrightness and homogeneity make them the best measure of distance, and hence dark energy, in the Universe

Supernova Legacy Survey (2003-2008)Megaprime5 year survey, goal: 500 distant SNe Ia to measure wUses CFHT/Megacam36 CCDs, good blue response4 filters for good k-corrections and color measurement

CFHT-LS OrganisationSNLS collaborationData-processing

Major Spectroscopic ProgramGemini (Canada/UK/USA)120 hrs/yr (60:40:20)VLT (France/Other Euros)120 hrs/yrKeck (through LBL)40 hrs/yrCosmological analysesMagellan near-IR study (Freedman et al.)Rest-frame I-band Hubble diagramKeck SN Ia UV study (Ellis/Sullivan et al.)LRIS high-S/N - metallicity through UV linesTesting accuracy of k-corrections in the UVSN IIP study (Nugent/Sullivan/Ellis et al.)Using SNe IIP as standard candlesIndependent Hubble diagram to z=0.5

CFHT-LS (imaging) 2003-2008DEEPWIDEGalaxy studiesTime sequenced dataset (202n over 5 years)Cosmic shearClusters

Supernova Legacy SurveyKeck (8 nights/yr)Gemini N & S (120 hr/yr)VLT (120 hr/yr)Magellan (15 nights/yr)Imaging

Distances fromlight-curvesSpectroscopy

Redshifts Distances from cosmological modelDiscoveriesLightcurvesgriz every 4 days during dark time

k-corrections in SNLS

k-corrections in SNLS

Making a standard candle1. Phillips relation: A correction to SN Ia light-curves based on light-curve shape drastically improves the quality of the standard candle.Time Brightness Brightness Time 56Ni 56Co 56Fe powers the SN Ia light-curveConventional Wisdom:SNe are a one-parameter family defined by amount of 56Ni synthesized in the explosion. More 56Ni greater luminosity higher Temperatures higher opacity broader LC

Colour at peakMaking a standard candle1. Phillips relation: A correction to SN Ia light-curves based on light-curve shape drastically improves the quality of the standard candle.2. SN colour: A correction to the SN luminosity based on the SN colourBlueRedFainter

Making a standard candle1. Phillips relation: A correction to SN Ia light-curves based on light-curve shape drastically improves the quality of the standard candle.Brightness Brightness Time 20%7%!Many methods:Stretch Perlmutter 97, 99(M)LCS(2k2) Riess, 95,96, Jha 07SALT(2) Guy 05, 07SiFTO Sullivan 07CMAGIC Wang et al.; Conley 06m15 Phillips 93; Hamuy 95; Prieto 062. SN colour: A correction to the SN luminosity based on the SN colour

Local SN Ia Hubble DiagramsJha et al. 2007Prieto et al. 2006Most light-curve fitting techniques fare equally well

Light-curve fit parameters from different fitters are tightly correlated

A Typical SN

What we need to measurePeak brightnessLightcurve width (stretch)Colour (c)

SNLS: Current statusSurvey running for 3.5 years~310 confirmed distant SNe Ia (+ 40-50 not yet processed)~ Largest single telescope sample of SNeOn track for 500 spectroscopically confirmed SNe Ia by survey end (>1000/>2000 total SNIa/All SN light-curves)

Rolling light-curves

SNLS 1st yearM = 0.263 0.042 (stat) 0.032 (sys)=-1.02 0.09 (stat) 0.054 (sys)Astier et al. 2006349 citations(187 in refereed journals)

Third year SNLS Hubble Diagram (preliminary)Preliminary3/5 years of SNLS~240 distant SNe IaIndependent analysis to 1st year:Different calibration routeDifferent photometric methodsDifferent SN light-curve analysis toolsSullivan et al. 2007

Third year SNLS Hubble Diagram (preliminary)M=0.3, =0M=1.0, =0Best-fit for SNLS+flatnessPreliminary(error was 0.042 in A06)Sullivan et al. 2007

Cosmological Constraints (Preliminary)SNLS+BAO (No flatness)SNLS + BAO + simple WMAP + FlatBAOBAOSNeSNeWMAP-36-7% measure of Sullivan et al. 2007

Future Prospects with SNLSCurrent constraints on : =-1 to ~6-7% (stat)

>-0.8 excluded at 3-sigma level

At survey end a 4-5% statistical measure will be achieved:500 SNLS + 200(?) SDSS + new local samplesImproved external constraints (BAO, WMAP, WL)

Systematic errors becoming ever more important

Potential SN Systematics in measuring w(a)Experimental SystematicsCalibration, photometry, Malmquist-type effects

Contamination by non-SNe IaMinimized by spectroscopic confirmation

K-correctionsUV uncertain; golden redshifts; spectral evolution?

Non-SNe systematicsPeculiar velocities; Hubble Bubble; Weak lensingExtinctionEffective RB; Dust evolution

Redshift evolution in the mix of SNePopulation drift environment?

Evolution in SN propertiesLight-curves/Colors/LuminositiesMore mundaneMore scientifically interesting

Hubble BubbleLatest MLCS2k2 paper (Jha 2007)MLCS2k2 attempts to separate intrinsic colour-luminosity and reddening

3 decrease in Hubble constant at 7400 km/sec local value of H0 high; distant SNe too faint

Local void in mass density?

Could have significant effects on w measurementMLCS2k2SALTNo Bubble with other light-curve fitters!Conley et al. (2007)

Light-curve fit parameters from different fitters are tightly correlated

Handling colour in SN IaColour is the most important correction to SN Ia luminositiesUnderlying physics: Redder SNe Ia are fainter due toExtinction along the line of sightIntrinsic luminosity/colour relationship of the SN population

Two basic approaches:Attempt to identify intrinsic relationship and assume standard dust dominates the rest (Jha et al.)

Fit a luminosity/colour relationship empirically on the SN data=4.1

Bubble significance versus Conley et al. (2007)

All fitters agree:

What does =2 (RV=1) mean?Very strange dust? But RV=1 is not seen anywhere in the Milky WayDust around SNe is changed by the explosion?

Most likely: SNe have an (as yet) uncorrected-for intrinsic colour-luminosity relationship

While fitting empirically for may be empirical, currently its the best way

Potential SN Systematics in measuring w(a)Experimental SystematicsCalibration, photometry, Malmquist-type effects

Contamination by non-SNe IaMinimized by spectroscopic confirmation

K-correctionsUV uncertain; golden redshifts; spectral evolution?

Non-SNe systematicsPeculiar velocities; Hubble Bubble; Weak lensingExtinctionEffective RB; Dust evolution

Redshift evolution in the mix of SNePopulation drift environment?

Evolution in SN propertiesLight-curves/Colors/LuminositiesMore mundaneMore scientifically interesting

Potential SN Systematics in measuring w(a)Experimental SystematicsCalibration, photometry, Malmquist-type effects

Contamination by non-SNe IaMinimized by spectroscopic confirmation

K-correctionsUV uncertain; golden redshifts; spectral evolution?

Non-SNe systematicsPeculiar velocities; Hubble Bubble; Weak lensingExtinctionEffective RB; Dust evolution

Redshift evolution in the mix of SNePopulation drift environment?

Evolution in SN propertiesLight-curves/Colors/LuminositiesPopulation Evolution

?White DwarfMany uncertainties: Nature of progenitor system the second star Single versus double degenerate? Young versus old progenitor? Explosion mechanism? Effect of progenitor metallicity on luminosity?

Host galaxies impact SN propertiesSome evidence that SNe Ia in ellipticals show smaller scatterSullivan et al. (2003)e.g. Hamuy et al. (2000)SN Ia Light-curve shape depends on morphologyLow stretchHigh stretch

PassiveStar-formingStarbursting Little morphological information available

CFHT u*griz imaging via the Legacy program.

PEGASE2 is used to fit SED templates to the optical data.

Recent star-formation rate, total stellar mass, mean age are estimated.

Hosts classified according to physical parameters instead of what they look like.Typing of SNLS SN Ia hostsSullivan et al. (2006)

SNLS: SN rate as a function of sSFRSN Ia hosts classified by star-formation activity

Per unit stellar mass, SNe are at least an order of magnitude more common in more vigorously star-forming galaxiesSNLS passive galaxies

SNLS selection of hostsD2 ACS imaging

Plenty of irregular/late-type systems

Few genuine ellipiticals

SN Ia Stretch dependencies170 SNe Ia(Update from Sullivan et al. 2006; better zeropoints, host photometry, more SNe)PassiveStar-formingStretch versus mean ageStretch by galaxy star-formation activityThe majority of SN Ia come from young stellar populations

Recent SNLS evidence for two components for SNe IaOlder progenitor SNe

Empirically, SN Rate is proportional to galaxy mass

Preferentially found in old stellar environmentsTypically fainter with faster light-curves (low stretch)Younger progenitor SNe

Empirically, SN rate is proportional to galaxy star formation rate

Exclusively found in later type star-forming galaxiesTypically brighter with slower light-curves (high stretch)(Extreme example: SNLS-03D3bb?)

Recent SNLS evidence for two components for SNe IaOlder progenitor SNe

Empirically, SN Rate is proportional to galaxy mass

Preferentially found in old stellar environmentsTypically fainter with faster light-curves (low stretch)Younger progenitor SNe

Empirically, SN rate is proportional to galaxy star formation rate

Exclusively found in later type star-forming galaxiesTypically brighter with slower light-curves (high stretch)(Extreme example: SNLS-03D3bb?)Could evolution between the two components with redshift distort the dark energy signal?

SN population drift?Relative mix of evolves with redshift

A+B predictions, but similar for any two component model

Sullivan et al. 2006

Stretch versus redshift (3rd year)?

Evolution in Stretch?

Gaussians predicted evolution from A+B model

Average stretch, and thus average intrinsic brightness of SNe Ia evolves with redshift

but

if stretch correction works perfectly, this should not affect cosmologyHowell et al. 2007Nearbyz1

Effect on cosmologyExtreme case using SNLS 1st year

Use only s0.4

Effects of evolution smaller than error budget for determination of , must be studied closely to determine the effect on measuring w(a)SNLS yr 1 N: 115 = 1.6 = 1.8w = -1.05 0.09M = 0.27 0.02Yr 1 s split N: 56 = 1.4 = 1.8w = -0.92 0.15M = 0.28 0.03

Split by s (preliminary)

SM0.95

Stretch correction across environmentsConley et al. 2006No evidence for gross differences between light-curves in passive and active galaxiesRest-frame B composite light-curve

Morphological Hubble diagram

Split by host galaxySN SubsetsProblems: Low-redshift sample very small, Malmquist correction likely to be differentPassiveStar-forming=1.34 0.25=2.52 0.2~0.10 mag=-0.880.11=1.19 0.15=2.71 0.2~0.140 mag =-1.040.08Preliminary

Do SNe Ia Evolve? UV Spectrum Probes MetallicityLentz et al. (2000)

High S/N spectra of SNLS SNeEllis, Sullivan et al. 2007

Light-curve width dependenceImplications for JDEM: k-corrections must improve in accuracy

SummarySNLS is a well-controlled, calibrated and understood experiment

Current SN Ia measurement determine to ~ 6-7%SNLS is the best cosmological SN Ia dataset available

In 2-3 years, 4-5% will be achieved

Colour is the critical correction to SN Ia distances; currently can only be handled empirically

SNe Ia have a range of progenitor agesImpacts on light-curve shape: faster/older & slower/younger

SNe in passive galaxies are better standard candles?More homogeneous stellar population? Less dust?

Point is: look at diversity of host types!